What is a Recommender System?

Recommender systems are used in every day life - to suggest a movie or what you should buy next. What are they really and how do they work?

Recommender Systems are a very common type of Machine Learning algorithm (in case you need them, check out our our resources in this series - What is Machine Learning? and What is an algorithm?)

Let's think about this with a really well known example - a recommender system. Most people will have used some shopping website (e.g. Amazon or ASOS) or watched a streaming service (e.g. Netflix or Disney+). Often these services have a recommended for you section or something similar. A computer programme (called an algorithm) is looking at what you have previously watched or bought, and is guessing what you will want to watch or buy next. Now for the learning part, based on whether you watch or buy the items recommended by the system, the algorithm will use this information to recommend or not recommend this to people in the future (including you). Therefore the algorithm learns from new data being added to the system.

A diagram showing how a recommender system works with Netflix as an example

Not all recommender systems work - have you ever bought an item on Amazon, say a hairdryer. Then Amazon keeps recommending hairdryers, like you collect them and not like the true version where you only need one (unless I'm sorely wrong and everyone who buys a hairdryer buys multiple).

Checkout our other resources:

What is AI?